A Broad Discussion of Speakers with Major Audio Luminaries

[dfuller]

Thought experiment:
An anechoic chamber is a tool.
Which problem is it better suited to solve, the problem of the loudspeaker, or the problem of the room?

It solves the problem of the room corrupting the measurement of the speaker (well, mostly).

 

[runsWithStupid]

Thought experiment:
An anechoic chamber is a tool.
Which problem is it better suited to solve, the problem of the loudspeaker, or the problem of the room?

It’s not meant to solve either. The tool historically removed the need for a large open field to take measurements. So weather you use a Klippel, a chamber or an open field those tools solve the reflection problem when taking measurements.

 

[Vergiliusma]

Well, it is to solve issues with the speakers. If you're a designer and you want to know what the loudspeaker by itself is doing, not the loudspeakers and the room, you need to remove its effects.

Trying to apply EQ from room measurements above the transition frequency typically doeasn't end well.

Remember your Toole, Floyd that is.

 

[runsWithStupid]

something that some designers will hate but others will love is AI will eventually be able to take all the CAD data and schematics along with spinorama data and actively propose design fixes. Some designers will embrace it others will be vehemently opposed to its use. I’m already using it to check drawings and it’s went from almost useless to okay at finding obvious stuff that I miss.

 

[Sigberg Audio]

Since I'm already on a roll hurting feelings, I might as well keep going.

Speakers as a collective whole from just a decade ago compared to speakers of today has a noticeable objective performance gap.

The best speakers of yesterday were all designed using an anechoic chamber.

I believe the gap between the performance of speakers from yesterday and today is largely due to the tooling. The best method of yesterday was an anechoic chamber, but it was super laborious. You had to manually position the mic for each measurement. Then you have to wait for a day of moderate temperature, humidity and wind to take the speaker outside to measure the bass on a 20-30 foot pole and then stitch it together with the initials anechoic measurements. Maybe some had custom toolings that would reduce the time and effort in measuring speakers.

I once tried to do a quasi-anechoic measurement manually just to go through the experience, it was such a waste of my time and never attempted it again, my time is way too valuable for this ancient practice. It's akin to my most recent experience with coding and debugging without having AI to support you.

But today, the NFS is a fully automated robotic arm, you set it up, click the go button and walk away for few hours and you come back with high resolution, high precision results. It makes iterations easy and reliable. Other tools like the Dynaduio Jupiter enables faster measurements as well.

Since I assume you are okay with direct feedback given your own directness, I think this is flat out wrong.

Almost no one has a Klippel NFS, so it would be strange if that is the main reason speakers has improved the last decade. To my knowledge for instance Dutch&Dutch does not have an NFS, to pick just one speaker that is universally considered objectively good on this forum. Does Revel have one? I don't know. Kef? Not to my knowledge.

Since you have pointed out that you are careful with your words, I see that you don't explicitly mention the Klippel NFS as that better tooling you are talking about, but it seems implied in your text.

But I would agree that something as "simple" as REW is probably used extensively by many speaker manufacturers to great effect, and it is something not available until relatively recently. But it doesn't take away the labour of measurements.

One the point of speed; For development iterations you can actually take manual measurements way faster than waiting for full NFS measurements if you already know the speaker and how it measures. So the NFS would not be my tool of choice for fast iterations, rather to verify the results via a full spinorama now and then. Currently I use an anechoic chamber for that, having an NFS in an adjacent room would surely be a bit more effective for that particular task.

But a good anechoic chamber is certainly accurate enough to design SOTA speakers. And we have evidence of great measuring speakers even from "horrible" manufacturers like B&W from back in the 90s. So it has certainly been doable for a long time, no Klippel NFS required.

EDIT: Note that this is not an attempt to bash the Klippel NFS, I would love to have one. Just saying that since almost no one has it, it can't reasonably the source of improved loudspeaker designs. Possibly indirectly in the sense that its existence has increased focus on measurements.

 

[Keith_W]

Almost no one has a Klippel NFS, so it would be strange if that is the main reason speakers has improved the last decade. To my knowledge for instance Dutch&Dutch does not have an NFS, to pick just one speaker that is universally considered objectively good on this forum. Does Revel have one? I don't know. Kef? Not to my knowledge.

I can tell you for sure that KEF does not have a Klippel. I saw a factory video of KEF some time back and they showed their very small anechoic chamber. They said they were fortunate to have a huge parking lot which is where they do their ground plane measurements, but given that it's England they had to wait for the right time of the year.

Anyway I agree with you. You don't need a Klippel. Anybody who is determined enough can obtain the information that they need. The cost is time and effort, rather than money.

 

[Mart68]

Tannoy also used their parking lot and a crane.

I remember years ago one of their engineers telling a story where they were measuring a big PA speaker out there and a school that was a mile away called to complain about the noise.

 

[PristineSound]

Since you have pointed out that you are careful with your words, I see that you don't explicitly mention the Klippel NFS as that better tooling you are talking about, but it seems implied in your text.

Very careful with choosing my words, especially on the Internet:

Other tools like the Dynaduio Jupiter enables faster measurements as well.

The NFS is one of the tools. I don't believe you must do a full scan with the NFS each time. KEF has a custom microphone array with custom software they use in their anechoic chamber. So one measurement captures multiple positions. That is tooling.

Also I want to add, by me saying the best method of yesterday, does not imply it's a bad method today. Specifically using an anechoic chamber.

 

[PristineSound]

But a good anechoic chamber is certainly accurate enough to design SOTA speakers. And we have evidence of great measuring speakers even from "horrible" manufacturers like B&W from back in the 90s. So it has certainly been doable for a long time, no Klippel NFS required.

I also want to add that when I say tooling, I don't just mean measurement tooling. For instance, FEA with Comsol, where the acoustic module didn't become available since mid-2000's. It enables faster prototyping, companies like Perlisten was able to use Comsol to POC on a computer first and it helps them understand design behaviors as often times complex physics isn't always clear how it would behave. Other tools include vibrometers, which I believe Ascend also used to measure cabinet resonance.

I believe it is the availability and the adoption of better tools that enables better speakers today. I would hope most people on ASR would agree that speakers of today mops the floor of speakers of yesterday, at least from an objective performance perspective.

You have a different opinion on why speakers have gotten better, you don't seem to believe it has much to do with tooling. In such case, I would like to hear your opinion as to what you believe the key factors today that makes speakers better than of those from yesterday.

 

[Mart68]

I would hope most people on ASR would agree that speakers of today mops the floor of speakers of yesterday, at least from an objective performance perspective.

'Mops the floor' - no I wouldn't agree. Incrementally better than the best from the late '70s on, then yes.

 

[PristineSound]

'Mops the floor' - no I wouldn't agree. Incrementally better than the best from the late '70s on, then yes.

When I see spins like these in modern day speakers relative to the best of the best in the 70's, 80's and even 90's, it does make me want to use the term "mops the floor".

 

[sam08861]

I'd also wonder if the 'objective performance' is the difference in the historical process and preference of voicing vs. the contemporary desire (or obsession) of select groups for a ruler flat FR by many who seek to emulate studio monitors in their home systems.

In modern stuff with DSP or complex circuits and individual amplifiers per driver, with active crossover, this can be very flat as the number EQ filters is practically endless in the digital domain. The advancements are in the digital domain, not the bi-tri etc. amping, as that's been standard practice for live venues since the 60s-70s.

Some people also prefer a 'straight wire with gain' approach and look to minimizing filters, avoidable adc/dac or complex circuits and IC's in the active xo to speaker domain.

Many people prefer a psychoacoustically aligned FR such as a Harman curve, particularly for low level listening and want to minimize having to have additional EQ from flat in the signal chain or the unnecessary circuitry in the xo/eq to make it flat. I'm in that camp and would prefer to EQ once from the speakers natural response. Whether this happens before or after the cross is also a design consideration.

Other than flat FR, I don't feel there's much material difference between the designs or yesterday and today. No real advancements in material science. Things like beryllium, aren't new and exotic cone materials, have advantages and disadvantages vs. paper, but all materials are compromised in one way or another and a heck of a lot of high end speakers still use paper, copper coils and aluminum baskets, just like they did in the 60s.

I think measurements are certainly an evolution in speaker design tools and many folks needing that confirmed ruler flat studio monitor sound signature for editing.

However, I'd venture to say one persons objective performance is different from anothers, depending on the goals and design criteria. The listening enjoyment is a goal for my home system, whereas this wouldn't be at all the case for a studio, needing that flat FR.

 

[Sigberg Audio]

You have a different opinion on why speakers have gotten better, you don't seem to believe it has much to do with tooling. In such case, I would like to hear your opinion as to what you believe the key factors today that makes speakers better than of those from yesterday.

I commented explicitly about the Klippel NFS, since again that seemed quite heavily implied in your post. If you were referring to tooling in general and not Klippel NFS specifically, I would agree that it's at least part of the equation.

The other major component is that information is more readily available and accessible (aka internet). This makes it much easier to learn / educate yourself on just about anything (not only loudspeaker design).

 

[amirm]

Does Revel have one?

Of course they do. I have been to one of them but believe they have two.

 

[Ropeburn]

Fun fact: Heco has been using it for years as well. Certainly helped with designing their nice directivity and low distortion speakers, the drivers, and predicting how they work in average rooms.

 

[PristineSound]

I commented explicitly about the Klippel NFS, since again that seemed quite heavily implied in your post. If you were referring to tooling in general and not Klippel NFS specifically, I would agree that it's at least part of the equation.

The other major component is that information is more readily available and accessible (aka internet). This makes it much easier to learn / educate yourself on just about anything (not only loudspeaker design).

Yes, I was referring to tools as a whole:

I believe the gap between the performance of speakers from yesterday and today is largely due to the tooling.

This includes but not limited to acoustic and vibration measurement tools from Klippel or other manufactures, modeling software, custom built tools, heck even something as simple as laser distance meter (wasn't commercially available until the early 90's). In fact, I am almost certain, in the next few years, AI will be used to help with speaker design; if I had to bet money, I would say Comsol may embed AI into their software to help solve complex physics problems.

I expect speakers in another one to two decades would mop the floor of today's best speakers.

 

[a4eaudio]

But I would agree that something as "simple" as REW is probably used extensively by many speaker manufacturers to great effect, and it is something not available until relatively recently

For sure, relatively recently compared to the 1960s and 1970s. While REW was introduced in 2003, I don't think it offered a lot of the speaker-measurement stuff that we currently take advantage of until later. ARTA came out in 2006. LEAP/LMS came out in the mid-1980's. There was some DOS based software and punchcards even earlier that I have seen mentioned by some people, but I don't remember the name.

 

[RexrothPigeon]

I would hope most people on ASR would agree that speakers of today mops the floor of speakers of yesterday, at least from an objective performance perspective.

100% no doubt about it

 

[PristineSound]

Other than flat FR, I don't feel there's much material difference between the designs or yesterday and today. No real advancements in material science. Things like beryllium, aren't new and exotic cone materials, have advantages and disadvantages vs. paper, but all materials are compromised in one way or another and a heck of a lot of high end speakers still use paper, copper coils and aluminum baskets, just like they did in the 60s.

I disagree. Material is one thing, but there are advanced applications of the materials and advances in manufacturing.

Then have better methods and modeling.

At the end of the day, there are much more than FR: directivity, distortion, resonance.

I do agree at the end of the day, speakers gets "voiced" and that is a design decision.

 

[Mike123]

Fun fact: Heco has been using it for years as well. Certainly helped with designing their nice directivity and low distortion speakers, the drivers, and predicting how they work in average rooms.

At one point, I was very surprised how Heco produced excellent results even on budget speakers. Now I understand why. Using automated tools doesn't make things better and better. It pushes guaranteed good quality down to the mid- and low-end price segments and makes everything sold sound good, not just six-figure masterpieces.